A plasma is a special aggregate state, which is produced from a gas. Each gas essentially comprises atoms and/or molecules. In the case of a plasma, this gas is for the most part ionized. This means that by supplying energy the atoms or molecules are split into positive and negative charge carriers, that is, into ions and electrons. A plasma is suitable for the processing of workpieces, since the electrically charged particles are chemically extremely reactive and can also be influenced by electrical fields. The charged particles can be accelerated by means of an electrical field onto a workpiece, where, on impact, they are able to extract individual atoms from the workpiece. The separated atoms can be removed via a gas flow (etching), or deposited as a coating on other workpieces (thin-film production). Such processing by means of a plasma is used above all when extremely thin layers, in particular in the region of a few atom layers, are to be processed. Typical applications are semiconductor technology (coating, etching etc.), flat screens (similar to semiconductor technology), solar cells (similar to semiconductor technology), architectural glass coating (thermal protection, glare protection, etc.), memory media (CDs, DVDs, hard drives), decorative layers (colored glasses etc.) and tool hardening. These applications make great demands on accuracy and process stability. Furthermore, a plasma can also be used to excite lasers, in particular gas lasers.
To generate a plasma from a gas, energy has to be supplied to the gas. This can be effected in different ways, for example, via light, heat, and/or electrical energy. A plasma for processing workpieces is typically ignited and maintained in a plasma chamber. To that end, normally a noble gas, e.g. argon, is introduced into the plasma chamber at low pressure. The gas is exposed to an electric field via electrodes and/or antennas. A plasma is generated and is ignited when several conditions are satisfied. First of all, a small number of free charge carriers must be present, such that the free electrons that for the most part are always present to a very small extent are used. The free charge carriers are so forcefully accelerated by the electrical field that as they collide with atoms or molecules of the noble gas they release additional electrons, thereby producing positively charged ions and additional negatively charged electrons. The additional free charge carriers are in turn accelerated and, as they collide, generate further ions and electrons. An avalanche effect commences. The discharges that occur as these particles collide with the wall of the plasma chamber or other objects and the natural recombination of the particles counteract the continuous generation of ions and electrons, i.e., electrons are attracted by ions and recombine to form electrically neutral atoms or molecules. For that reason, an ignited plasma must be constantly supplied with energy in order for it to be maintained.
The supply of energy can be effected via a dc supply device or an ac supply device. The following remarks relate to ac supply devices for high frequency (HF) with an output frequency of >3 MHz.
Plasmas have a very dynamic impedance, which renders the design of the desired uniform HF power supply difficult. For instance, during the ignition process the impedance changes very quickly from a high value to a low value. Negative effective resistances can occur during operation, which reduce the current flow as the voltage rises, and undesirable local discharges (arcs) may occur, which may damage the material to be processed, the plasma chamber or the electrodes.
Power supply devices for plasmas (plasma supply devices) must therefore be designed for a high output power and a high reflected power. EP 1 701 376 A1 has shown that such plasma supply devices can advantageously be achieved by relatively small high frequency amplifiers, the output powers of which are coupled by a coupler, preferably by a 3-dB coupler (e.g., hybrid coupler, Lange coupler, etc.). For that purpose, the two high frequency amplifiers are connected to two ports of the hybrid coupler, hereafter called port 1 and port 2. The high frequency amplifiers are driven in such a way that their high frequency signals of the same fundamental frequency have a phase shift of 90° with respect to one another. At a third port of the hybrid coupler the first of the two high frequency signals is present lagging by 45°, and the second of the two high frequency signals is present leading by 45°. At a fourth port of the hybrid coupler the first of the two high frequency signals is present leading by 45° and the second lagging by 45°. By phase shifting of the two generated high frequency signals by 90°, these add up at the third port by constructive superposition, whereas at the fourth port they cancel each other out (destructive superposition). The high frequency amplifiers upstream of the coupler thus each require only half the power of the required high frequency output signal. These coupler stages can be cascaded to enable the use of high frequency amplifiers with even less source power or to achieve an even higher power of the high frequency output signal.
The fourth port of the hybrid coupler is normally terminated with a terminating resistance of the system impedance (often 50Ω). As described in EP 1 701 376 A1, a high frequency signal is expected at this port only when a high frequency signal reflected by the plasma load is in turn reflected at the high frequency amplifiers.
In the case of mismatching due to different impedances of plasma supply device and plasma load, the power delivered by the plasma supply device is partially or fully reflected. An impedance matching circuit (matchbox) can transform the impedance of the plasma load in certain ranges and match it to the output impedance of the plasma supply device. If the transformation range of the matching circuit is exceeded, or if regulation of the impedance matching circuit cannot follow a rapid impedance change of the plasma, then the total power delivered by the plasma supply device is not absorbed in the plasma, but rather reflection occurs again.
A high frequency signal reflected by the plasma load runs via an optionally present matching circuit back to port 3 of the hybrid coupler where it is split into two parts and retransmitted via ports 1 and 2 towards the high frequency amplifiers of the plasma supply device. In the process, the two parts of the reflected high frequency signal again experience an equal phase delay by 45° en route from port 3 to port 1 and an equal phase lead by 45° en route from port 3 to port 2. The result being that, at both outputs of the two high frequency amplifiers, the forward and reflected high frequency signals are superimposed differently by 180°. If, for example, a maximum constructive superposition of forward and reflected high frequency signal takes place at the output of the first high frequency amplifier, then this superposition will have maximum destruction at the output of the second high frequency amplifier. If the superposition at the first high frequency amplifier is such that the current maximum is ahead of the voltage maximum in time, i.e. the high frequency amplifier sees a capacitive impedance as load impedance, then the voltage maximum will be ahead of the current maximum in time at the second high frequency amplifier, i.e. the second high frequency amplifier sees an inductive load impedance.
The two high frequency amplifiers can be driven, for example, by a common drive signal generator. This drive signal generator can have a phase shift of 90° for the high frequency control signals at its two outputs. The two high frequency amplifiers can also be driven by a common high frequency driver transmitter, the output signal of which is split, for example, by means of a second hybrid coupler.
Plasma supply devices having two high frequency amplifiers can also be constructed with couplers having other phase shift characteristics. The above-mentioned properties/advantages are only achieved, however, when the phase shift is 90° or close to that. A phase-shifting coupler can also be achieved, for example, by means of a 0°-coupler (Wilkinson coupler) and corresponding phase lag elements, for example, phasing lines.
One of the most common circuit variants of high frequency amplifiers in plasma current supplies is a class D amplifier with a switching bridge. A switching bridge has at least two switching elements, such as e.g. MOSFETs, which are connected in series; the junction of the switching elements represents the midpoint of the switching bridge. The midpoint of the bridge arm is connected alternately to the positive or negative pole of a power D.C. supply by the two switch elements (hereafter also called switching elements or switches). The alternating control of the two switch elements and of the switch elements of any second bridge arm present is effected by the drive signal generator, which can contain an oscillator, which determines the frequency of the output signal, and further components, such as inverters, phase shifters and signal formers.
A full bridge circuit consists of two bridge arms (half-bridges), the midpoints of which are connected at the desired frequency in each case inversely to the positive and negative pole of the D.C. supply. The alternating current load is arranged between these two midpoints. An additional capacitor to free the output signal from a D.C. offset is unnecessary.
To avoid switching losses, at the time the individual switch elements of a full bridge are switched on, there should be no appreciable voltage difference between the two power electrodes (generally drain and source of the MOSFET in question). This switching behavior is called zero voltage switching. This is achieved by the switching bridge operating on a load impedance having an inductive character. This means that the switching bridge, i.e. the high frequency amplifier, sees an inductively absorptive load impedance. The self-induction of a primary winding of a power transformer that is connected to the midpoint of the switching bridge can be used for that purpose. A voltage is induced upon the initially one-sided interruption of the current flow through the primary winding. When the components are of suitable dimensions and allowances are made for their parasitic properties, and with the correct choice of switching/delay time, the potential at this end of the primary winding which is momentarily disconnected from the power D.C. supply is just as high as the potential at the particular connection of the direct voltage source that is now to be connected to this end of the primary winding with this half-bridge.
In contrast, a load impedance of a capacitive character (the HF amplifier sees a capacitively absorptive load impedance) is unfavorable for the switching bridge, as the midpoint retains its previous potential during the switching-over procedure and thus, a voltage difference up to the voltage of the power D.C. supply is present at the element now switching on.
Under certain reflection conditions, it may even happen that one of the high frequency amplifiers supplies power to the other; the load impedance seen by this high frequency amplifier to which power is supplied has a negative effective resistance, and hence the output power of this high frequency amplifier becomes negative. The high frequency amplifier is, as it were, “fed.” This means that the “fed” high frequency amplifier sees an emissive load impedance, which can be inductive or capacitive.
Depending on the operating states and the specifications of the components of the high frequency amplifiers, that is, how close to the specification limit these are operated with the supply voltage, how good the cooling is, how quickly the normal operation (switching operation) of the switching elements of the HF amplifier can be terminated/interrupted, how well zero voltage switching has to be realized, an admissible operating range arises. If therefore the load impedance lies in an operating range dependent on the component specifications and operating states of the HF amplifiers, the HF amplifiers can be operated without fear of damage or destruction. At the same time, certain (inductive) emissive load impedances may also be admissible. Load impedances outside the admissible load impedance range should be avoided.